Tyre deformation or just cutting track?

[Dravonic]

So... something caught my eye when watching the gamersyde video and I would like to know what others think. I realized it when taking a closer look to determine if we have a case of flying tires or not (BTW, I think we do).

Take a look yourself:

Full:

Close up:

Half speed close up:

I have my doubts, but if I had to pick I would say it's just cutting the track

 

[No46_TheDoctor]

What do you mean "flying tyres" ?

 

[balang_479]

I think it could be a case of cutting track but acting as tire deformation... so the physics for tire deformation are there but they dont need to model accurate tyre deflection which could be quite demanding....

 

[CodeRedR51]

Those GIF's are also good for seeing just how bumpy the track is in that one spot.

 

[Stonemonkey]

Graphically it looks like it's just penetrating the track but when calculating collisions it can be fatal to let penetations occur, suddenly you have points on the wrong side of a surface (vertices of the tyre underneath the track surface) which totally messes things up and there's no way they would let that happen for physics calcs. Watch the shape of the track it's going over and the motion of the car, it looks like it could be modelled in the physics but not in the graphics, perhaps the graphics code/models being used in this aren't capable of displaying what's happening.

 

[Dravonic]

What do you mean "flying tyres" ?

I mean I think the car is moving around but the tires are not touching the road as they should. Appears to be happening sometimes. Not "hit a bump and fly for a moment". I mean run in the air instead of the road.

But that's not what this thread is about.

 

[fcautoX]

I'ts hard to say but i think its cutting the track. PD usually doesnt like to waste precious processing power to animate miniscule things such as visual tire deformation. The calculations are still there though, which is what i care about

 

[Owen.C93]

Cutting track most likely. GT5p let you put a big chunk of your rear end under the track.

You can do this by driving up the camber on Daytona.

 

[newbielives]

I love how the car bounces so realistically.
Like everyone said, its probably clipping and not tire deformation.

 

[Stonemonkey]

Cutting track most likely. GT5p let you put a big chunk of your rear end under the track.

You can do this by driving up the camber on Daytona.

That would be slightly different, In GT5P I doubt there are is any collision detectiion between the cary body and the track surface as it's not something that would generally occur (there will be some special cases) whereas the tyres are what is tested for collision with the track and is what matters most for physics.

 

[Dravonic]

Graphically it looks like it's just penetrating the track but when calculating collisions it can be fatal to let penetations occur, suddenly you have points on the wrong side of a surface (vertices of the tyre underneath the track surface) which totally messes things up and there's no way they would let that happen for physics calcs. Watch the shape of the track it's going over and the motion of the car, it looks like it could be modelled in the physics but not in the graphics, perhaps the graphics code/models being used in this aren't capable of displaying what's happening.

That's what I would bet on. There's 2 tracks, one to calculate the physics and another one to show at the screen and those 2 tracks don't perfectly match. That's causing some tires to fly and others to sunk in the track. This kind of thing is a quite common occurrence on games and that's probably what's happening.

It's just that it looks so convincing. Looks like it flattens when coming out of a bump. Really looks like it's deforming lol. I though I would ask to see what you guys think.

 

[Owen.C93]

That would be slightly different, In GT5P I doubt there are is any collision detectiion between the cary body and the track surface as it's not something that would generally occur (there will be some special cases) whereas the tyres are what is tested for collision with the track and is what matters most for physics.

Yeah your right. Should of thought of that.

 

[Owen.C93]

That's what I would bet on. There's 2 tracks, one to calculate the physics and another one to show at the screen and those 2 tracks don't perfectly match. That's causing some tires to fly and others to sunk in the track. This kind of think is a quite common occurrence on games and that's probably what's happening.

Why would they do that? surely if there's a bump in the real physics track they would add a bump on the graphic overlay to?

 

[Stonemonkey]

That's what I would bet on. There's 2 tracks, one to calculate the physics and another one to show at the screen and those 2 tracks don't perfectly match. That's causing some tires to fly and others to sunk in the track. This kind of thing is a quite common occurrence on games and that's probably what's happening.

It's just that it looks so convincing. Looks like it flattens when coming out of a bump. Really looks like it's deforming lol. I though I would ask to see what you guys think.

I'm not sure about the two tracks thing, the movement seems to be relative to the bumps that are visible in the closeup clips.
The tyre mesh which isn't used for physics calcs looks to me like it's sinking through the track surface as the physics calculates some tyre deformation.

 

[Dravonic]

Why would they do that? surely if there's a bump in the real physics track they would add a bump on the graphic overlay to?

I'm not quite sure how they are doing things nowadays, but I know that a few gens back you have separate assets for physics and graphics. That's just how it was(is?) done. Per polygon physics calculation is too demanding to the processor so they normally have a much simpler shape to calculate when you hit a wall and things like that.

For example, take a GT5P car with 200k polygons. To calculate if you crash another car you can use the 200k polygons, or a simple cube. It's much faster to calculate if you penetrated a cube than it is to calculate if you penetrated a 200k polygon car. In fact it may be still impossible to calculate such a thing in real time nowadays. That's why.

I'm not sure about the two tracks thing, the movement seems to be relative to the bumps that are visible in the closeup clips.
The tyre mesh which isn't used for physics calcs looks to me like it's sinking through the track surface as the physics calculates some tyre deformation.

I'm not sure if there's a need to use a simpler track to calculate the physics so I suppose you can be right. The physics calculates tire deformation and let the graphical tire sink in the track to give a visual effect. Sounds possible.

 

[Stonemonkey]

I'm not quite sure how they are doing things nowadays, but I know that a few gens back you have separate assets for physics and graphics. That's just how it was(is?) done. Per polygon physics calculation is too demanding to the processor so they normally have a much simpler shape to calculate when you hit a wall and things like that.

For example, take a GT5P car with 200k polygons. To calculate if you crash another car you can use the 200k polygons, or a simple cube. It's much faster to calculate if you penetrated a cube than it is to calculate if you penetrated a 200k polygon car. In fact it may be still impossible to calculate such a thing in real time nowadays. That's why.

It'll still be like that, I'd bet the collision detection for the tyre and track is just a vector on the wheel plane which is perpendicular the the line the wheel points along the track surface.

 

[Owen.C93]

Oh ok.

Maybe track has 30% movement and suspension has 70%, if you notice in the vid one wheel comes off the floor as the other goes into the ground. Maybe under lots of stress it's easier to give the suspension a smaller amount of give to keep consistency.

 

[happycorey]

I hope they do it, and do it right, and then just say it's in the game like it's not a big deal

 

[Stonemonkey]

I hope they do it, and do it right, and then just say it's in the game like it's not a big deal

I don't think it really is all that big a deal, lateral deformation would be.

 

[happycorey]

I'm just saying it because FM3 was like ZOMG WE GOT TIRE DEFORMATION

 

[Stonemonkey]

I'm just saying it because FM3 was like ZOMG WE GOT TIRE DEFORMATION

Lateral, LOL @ what's going on with the graphics in this though

http://www.youtube.com/watch?v=eDkMCA_wyRU

One thing to point out though, any computer calculated physics can not even be as accuate as the least accurate part of the system and a lot of physics in racing games is approximate or fudged so saying two games feature tyre deformation does not mean it's been implemented the same way or even if it has it might not have the same behaviour, it all comes down to the engineers and programmers who've worked on it.

 

[evilfoxhound]

That first gif...... Love to me.

 

[Jay]

One thing to point out though, any computer calculated physics can not even be as accuate as the least accurate part of the system and a lot of physics in racing games is approximate or fudged so saying two games feature tyre deformation does not mean it's been implemented the same way or even if it has it might not have the same behaviour, it all comes down to the engineers and programmers who've worked on it.

So very true. 👍

 

[LordVonPS3]

I want to make a point on this topic as there appears to be some falsely made assumptions...

It is completely inconsequential to render the scene to visibly show the car suspension in action and to visibly show tyre deformation. The car tyres do not need to "cut the track". Collision detection is done by calculating where vertices would intersect during the following frame. The car suspension physics couldn't work correctly if the tyres did cut the track as the suspension would not compress correctly.

You will note that the game is rendering the same number of vertices in the scene and there's absolutely no performance impact to the game's visuals just to do that rendering. It's nothing like having to render additional scenery / another car. *All* the hard work is being done by the GT physics engine and car modeling simulation. This is Polyphony Digital's showcase.

 

[Dravonic]

I want to make a point on this topic as there appears to be some falsely made assumptions...

It is completely inconsequential to render the scene to visibly show the car suspension in action and to visibly show tire deformation. The car tires do not need to "cut the track". Collision detection is done by calculating where vertices would intersect during the following frame. The car suspension physics couldn't work correctly if the tires did cut the track as the suspension would not compress correctly.

You will note that the game is rendering the same number of vertices in the scene and there's absolutely no performance impact to the game's visuals just to do that rendering. It's nothing like having to render additional scenery / another car. *All* the hard work is being done by the GT physics engine and car modeling simulation. This is Polyphony Digital's showcase.

I'm not sure I understand what you're trying to say but something is going on. Either the tire is deforming (I wouldn't count on it) or it's cutting the track. I believe it's only cutting the track visually due to inconsistencies in the track shown in the screen with the one used to calculate the physics.

 

[Darla Starch]

After watching the half-speed GIF a few times, I am under the impression that the tire is merely cutting the track...but then again, I guess we need things to talk about during the wait for the full GT5.

 

[Gaz Vonhammer]

Was just going to bring this up,I noticed when understeering though a corner and thought it was deformed,but also thought it could be going through the track.
looked convincing with the suspension movement.

 

[Leadfoot53]

looks like actual tire deformation IMO

 

[Cigol]

I'm fairly sure PC simulations have cited tyre clipping as a sort of feature in the past. It works as a compromise between deformation and no deformation. There's no visible deformation in GT5 (because it would be obvious if it were there) but there's definitely some clipping. Whether it's intentional or not is the question I guess.

 

[Vincent Viceroy]

Ha ha...looks quite convincing however I'd also agree that deformation is being calculated but not visually represented...so it's probably clipping.